Download VERY LARGE TELESCOPE

Transcript
VERY LARGE TELESCOPE
Meteorological
Prediction
Software
User and Maintenance Manual
Document No.:
VLT-TRE-ESO-17443-1678
Issue No.:
01.2
Date :
October 19, 1998
Prepared by:
P. Labit
Released by:
M. Sarazin
2
Meteorological Prediction Software: User and Maintenance Manual
IssueRev.
Date
SectionPage aected
ReasonInitiationDocumentRemarks
1.0
11.09.1998
1.1
06.10.1998
6.2.1
IR image transfer more robust satellite
1.2
19.10.1998
6.1.3
access to SQL database changed getlastmeteo
Change Record
3
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
Contents
1 Introduction
6
1.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Reference Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
6
8
User's Manual
9
2 ECMWF Products
9
I
2.1 Description of the products . . . . . . . . .
2.1.1 Archiving the products . . . . . . .
2.2 The ECMextrDAT program . . . . . . . . .
2.2.1 Conception of the program . . . . .
2.2.2 Use of the utility . . . . . . . . . . .
2.3 The GrADS front-end application . . . . . .
2.3.1 Introduction . . . . . . . . . . . . .
2.3.2 Execution of the GrADS application
2.3.3 Overview of the screen . . . . . . . .
2.3.4 Manual use of the Utility . . . . . .
2.4 Background applications . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3 Satellite products
3.1 Introduction . . . . . . . . . . . . . . . . . .
3.2 The McIdas IDL front end utility . . . . . .
3.2.1 The execution of this utility . . . . .
3.2.2 The manual use of the GView utility
3.3 Bakground applications . . . . . . . . . . .
9
9
11
11
11
13
13
13
13
14
15
17
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4 The Local products
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Background applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17
17
17
17
18
19
19
19
4
Meteorological Prediction Software: User and Maintenance Manual
5 The internet pages
5.1 Overview . . . . . . . . .
5.2 Web publishing scripts . .
5.2.1 ECMWF products
5.2.2 CIRA products . .
5.2.3 Verication pages .
20
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
20
20
20
21
21
6 Fortran programs used to produce the forecasts
23
II Maintenance manual
25
7 Maintenance of the scripts
25
7.1 Scripts related to ECMWF products . . . . . . . . . . . . . . . . . . . . . .
7.1.1 The transfer scripts ECMtoOlasg and ECMtoOlasg2 . . . . . . . .
7.1.2 The display scripts scrWeb0 and scrWeb4 . . . . . . . . . . . . . .
7.1.3 The verication scripts step1, step2, step3, getlastmeteo, step4.gp
7.1.4 The publication scripts publish0,publish4 . . . . . . . . . . . . . .
7.2 Scripts related to CIRA products . . . . . . . . . . . . . . . . . . . . . . . .
7.2.1 The satellite scripts satellite, goesGetSat . . . . . . . . . . . . . .
7.2.2 The goesGetSat script . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
8 The GrADS applications and utilities
38
8.1 The front-end utility mainMeteo.gs . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 The 'images' scripts script*.gs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.3 The verication scripts verif0.gs, writetxt.gs . . . . . . . . . . . . . . . . . . . . . . .
9 Maintenance of the Fortran programs
9.1 The ECMextrDAT utility . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2 The user-dened functions for GrADS . . . . . . . . . . . . . . . . . . . .
9.2.1 The bilinear interpolation biliLaSi, biliPara, biliChaj . . . . . . .
9.2.2 The cubic spline interpolation cuspLaSi, cuspPara and cuspChaj
9.2.3 The writPred and writPre2 functions . . . . . . . . . . . . . . . . .
25
25
26
28
33
34
34
37
38
38
42
45
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
45
46
46
47
47
5
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
10 Maintenance of other programs
48
10.1 The IDL programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2 The Java applets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A Appendix
A.1
A.2
A.3
A.4
A.5
48
48
50
Location of the Unix-scripts and their related les . . . .
Location of the GrADS applications and their related les
Location of the Fortran programs . . . . . . . . . . . . . .
Location of the IDL applications and their related les . .
Location of the java applets and their related les . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
B List of crontab les
50
52
54
55
56
57
List of Figures
1
2
3
4
5
6
7
8
Path followed by the data . . . . .
Title screen from mainMETEO.gs
The bottom button bar . . . . . .
The top button bar . . . . . . . . .
The Gview main button bar . . . .
The Gview display button bar . . .
The Satelitte applet . . . . . . . .
Principle of the Kalman lter . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
7
14
14
15
18
18
21
24
6
Meteorological Prediction Software: User and Maintenance Manual
1 Introduction
1.1
Scope
This document describes the software currently installed on the various ESO Workstations for the
purpose of producing forecasts of meteorological observing conditions at ESO Observatories
1.2
Description
The eciency of a ground based observatory is considerably increased if observations can be planned
in a exible way, taking into account the expected atmosphere environment of the night, in particular
with respect to standard meteorological quantities, cloudiness and water vapor.
After the positive conclusion of a feasibility study made by the CRS4 organisation CRS4 96
and
CRS4 97
under contract with ESO, it was decided to initiate a pre-operational phase. To do so,
data obtained from external providers have to be processed adequately to show that an operational
forecast for the observatories is possible. The aim of all scripts, applications and utilities presented
in this report is to publish and to archive data extracted from the dierent sources for the forecasts
of meteorological parameters for the sites of La Silla1 , Paranal 2 and Chajnantor3. To understand
properly the path followed by the data, a graphic is presented here, which will be used as a reference
all along the report.
We shall rst study how to use the dierent utilities developed to display or manipulate the data. The
second part is more detailed, with the description of all the tools developed and of their maintenance
which provide ESO with a functional and quasi-operational way to forecast the meteorological conditions above the three sites of interest. After a verication period of a few months, the system will
then integrate regular operations of the observatories.
1
located at 70W42',29S16'
located at 70W24',24S37'
3
located at 67W45',23S01'
2
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
Figure 1: Path followed by the data
7
8
Meteorological Prediction Software: User and Maintenance Manual
1.3 Reference Documents
-VLT-SPE-ESO-17410-1174
-VLT-TRE-CRS-17443-0001
-VLT-TRE-CRS-17443-0002
-VLT-TRE-ERA-17440-1038
-VLT-TRE-ERA-17440-1039
-VLT-TRE-UNI-17440-0004
-The Messenger No:89
VLT-ASM Seeing and Coherence Monitor, Dimm Upgrade Plan
V-2.3, Aug.98
Feasibility Study of a Meteorological Prediction Model for ESO
Observatories in Chile Phase A1-A2
11 Jul. 96
Feasibility Study of a Meteorological Prediction Model for ESO
Observatories in Chile Phase A3-A4
11 Apr. 97
Feasibility Study and Development of Software Computer Code
for Cloud Cover and Water Vapour Forecasts +appendix
Jan. 96
Feasibility Study and Development of Software Computer Code
for Cloud Cover and Water Vapour Forecasts. Phase II:
Parameter Denition and Requirements for an Operational System
27 Jul. 97
Feasibility Study of Seeing Prevision using the MESO-NH
meteorological model
Dec. 97
Predicting Observing Conditions at ESO Observatories - Reality
and Perspectives
Sep. 97
9
Part I
User's Manual
2 ECMWF Products
2.1 Description of the products
Every twelve hours, ESO receives from ECMWF4a set of les. These les arrive on the opus3 machine,
which is supposed to be one of the most reliable of the whole ESO network, for the user asmpredi
in the directory homeasmecmwf. They are all the same length, which is 34,816 bytes. In those
les we nd the value of 4 parameters temperature, u-component of wind, v-component of wind and
geopotential height for nine levels of pressure 100 mb, 200 mb, 300 mb, 400 mb, 500 mb, 700 mb,
850 mb, 925 mb and 1000 mb.
During the night, around 4:00 local time,ESO receives 7 les extracted from the 00Z dissemination
cycle of ECMWF. Those les cover a range of 36 hours of prediction, with time steps of six hours. It
is to say these are analysis for 00UT T + 0, and predictions for T+6, T+12, T+18, T+24, T+30,
T+36. There is one le per time step. The name of the les are ESDxxyyy0. The rst two numbers
xx represent the time step, with respect to the formula : xx = forecast time step 3 + 8. The three
following numbers yyy are the day of the year 000 to 366.
During the day, around 8:30 local time, ESO receives 20 les extracted from the 12Z dissemination
cycle of the ECMWF. Those les cover a range of 192 hours with a time step of 6 hours until T+36 and
12 hours after. It is to say we have the analysis for 12UT T + 0, and the forecasts for T+6, T+12,
T+18, T+24, T+30, T+36, T+48, T+60, T+72, T+84, T+96, T+108, T+120, T+132, T+144,
T+156, T+168, T+180, T+192. There is one le per time step, and the name of the les are
ESDxxyyy4. The rst two numbers xx represent the time step, with respect to the formula : xx =
forecast time step 6 + 4. The three following numbers yyy are the day of the year 000 to 366.
In each le we nd either the analysis or the forecast for three gridded region : the rst range from
65W,10S to 110W,45S with a step of 3 degree. The two others use a step of .5 degree, and range
respectively from 67W,22S to 71.5W,26.5S and from 67W,27S to 71.5W,31.5S.
These les respect the GRIB 92 ECMWF format ECMWF 97, and thus some utilities are needed to
decode them in order to get the data in a binary format readable by such utilities as GrADS 5 see
section 2.3.
2.1.1 Archiving the products
The rst thing to do when we receive the les is to archive them on the DMD archive machine olasg.
Two scripts were then written, which names are ECMtoOlasg and ECMtoOlasg2. They are located
in homeasmecmwf on opus3 and execute themselves once per hour. The aim of these scripts is in a
4
European Center for Medium range Weather Forecast, Shineld Park, Reading, Bershire, RG2 9AX, England. Web
site : http:www.ecmwf.int
5
GrADS, Grid Analysis and Display System, is a freeware. An homepage can be found at the URL:
http:grads.iges.orggradshead.html
10
Meteorological Prediction Software: User Manual
rst time to put the les on a big disk on olasg, and next to archive these les in a tar le on another
disk as backup of the rst one. The two scripts work exactly in the same manner.
To keep integrity, we transfer only entire leset. That is to say we wait either for the ESD20???0 le
or the ESD36???4 le before transferring the corresponding set. Then in a rst time, the script looks
for the last day of transfer from opus3 to olasg. In order to do this, we 'ftp' olasg and we look for
the last le transfered. Then we deduce the day of the last transfer. We increment this day variable,
and as long as there is a set of les to transfer from opus3, we transfer sets of le in the directories
home2asmprediECMWFdatadata00UT or ...data12UT according to the case on olasg. Next
we produce a tar le with the corresponding set of les, we compress this tar le and we transfer it
to the directory diskbasmprediECMWFarchive on olasg. We can then remove the les from the
opus3 machine.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
11
2.2 The ECMextrDAT program
2.2.1 Conception of the program
The data provided by ECMWF are in GRIB 92 ECMWF format ECMWF 97, which is not readable
by meteorological utilities like GrADS see section 2.3. Moreover, there are three grids mixed in the
les, and there is one le per time step, instead of one le per grid with all the times steps. Then we
need to convert the data in order to be able to visualize them.
ECMWF has developed some libraries of functions in fortran in order to decode this format of les. The
heart of the library is the function Gribex. To have explanations about this function and its coding,
please report to the le gribex.f, located on olasg in the directory diskb asmpredi ECMWF software
where the function and its parameters are explained ECMWF 95. In fact, this function decodes one
record of a GRIB le. The solution was then to use this function to sort the three types of grid, and
write, with one set of les from ECMWF, three binary les readable by most utilities.
The conversion utility, ECMextrDAT, is based upon a loop which length depend on the number of
les from ECMWF seven for 00Z or twenty for 12Z. In a rst step, it scans the GRIB les to sort
the records depending on the grid to which they belong. It creates the three tables, in which it puts
the exact place of the record in the original le.
In a second time, it calls a subroutine to sort the three arrays to respect the GrADS conventions:
the variables must be sorted rst by time, next by pressure level, and then by variable. Then the
subroutine returns the three sorted arrays to the main loop, in which the three arrays are read, and
the binary les written with respect to the place of the records in the arrays.
2.2.2 Use of the utility
The utility ECMextrDAT is located on the olasg machine in the directory home2 asmpredi ECMWF software.
It must be linked with the two libraries pbiolib.a and griblib.a.
The set of les to transform must be in the same directory as the utility. the command line is:
olasg: ECMextrDAT 125 0
The rst number is the day of the year of the les, and the second one is the extension of the les 0
for 00Z predictions and 4 for 12Z predictions.
Then ECMextrDAT reads the set of les from ECMWF and produces three or six binary les:
With the 00Z le set seven les, ECMextrDAT produces three les named ESDyyy0 x.dat
binary les and the three associated control les for GrADS. The extension yyy stands for the
day of the year, and x is either 1,2 or 3 1 stand for the general grid 3x3 degree, 2 is for the
Paranal-Chajnantor region and 3 means the the La Silla grid.
With the 12Z le set twenty les, the utility produces six les named ESDyyy4 x.dat binary
les and the six associated control les for GrADS. The extension yyy stands for the day of
the year, and x is between 1 and 6 1, 2 and 3 stands for the predictions with a time step of six
hours extracted from the ECMWF les with the same convention as above. 4, 5 and 6 stands
12
Meteorological Prediction Software: User Manual
for the long term predictions with a time step of twelve hours extracted from the ECMWF les
with 4 for the general grid, 5 for Paranal-Chajnantor and 6 for La Silla.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
13
2.3 The GrADS front-end application
2.3.1 Introduction
In order to visualize the data provided by ECMWF, an utility designed to display meteorological data,
GrADS, is used. Its aim is to display meteorological data. This utility is distributed as a freeware,
which means that it is free of charges. It was rst created at IGES by Mike Fiorino and Brian Doty. It is
permanently maintained and updated, and used by most meteorological institues worldwide. One can
visit the home-page for this utility which is located at the URL : http:grads.iges.orggradshead.html.
In order to import the data in GrADS, a transformation of the le was needed, which is done by the
previous utility see section 2.2
. But GrADS is not very user friendly. Then a tool was developed in
order to browse the data.
2.3.2 Execution of the GrADS application
The visualization script is located on the olasg machine, in the directory home2asmprediECMWFdatatmp.
Its name is mainMETEO.gs. It is adapted from a script developed by CRS4 CRS4 96, in order to
display the data sent by ECMWF.
To run the script, one just go into the previous directory and type 'grads'. This starts the GrADS
utility. After that, one just need to type 'run mainMETEO.gs' in order to execute the script. All the
transformation of the data les are done automatically by the application.
One will then be prompted for the day of the predictions one wants to examine. Then, the only
requirement is to enter a three digit code corresponding to the number of the day in the year. Then
one is prompted for the hour to examine 0 or 4, see above for details on the signicance of those
codes
. If 4 which means the 12Z predictions
is entered, the application will prompt for either the
short term predictions 0
or the long term ones 1
. If one entered 0 at the previous step, nothing is
to be done as only short term predictions are available.
If all the les exists and are correct, the script will perform a ECMextrDAT on the input les to
produce the three GrADS les needed plus the three .ctl, see section 2.2
. The user interface will
then be displayed in the graphic window.
2.3.3 Overview of the screen
When the scripts starts, a title screen is displayed and so are some buttons bars and menus. The
screen should look like the one presented on gure 2. There are menus and buttons on three side of
the screen. Below is a small explanation of the buttons, but a more precise explanation is in section
2.3.4.
On the top are the main tools of the utility zoom, type of projection, type of representation,
print or not,...
.
On the right are the buttons used to control the time and the pressure level.
On the bottom are the buttons used to control the type of data to display and the type of display.
14
Meteorological Prediction Software: User Manual
Figure 2: Title screen from mainMETEO.gs
The green button at the top left is the quit" button. When clicked, it stops the GrADS utility and
deletes all the temporary les created by the utility.
2.3.4 Manual use of the Utility
Types of data to display
The bottom button bar looks like the one presented on Figure 3. The
Figure 3: The bottom button bar
bottom button bar allows the user to choose the type of data to display. Since there are too many
types of data, those ones are split into two buttons bar. The swap between these two is done with the
bottom buttons '1' and '2'.
With the rst type of parameter, the geopotential height GH, the temperature T, the magnitude
of wind WIND mod, the direction of wind WIND vec and the streamlines of wind WIND str can
be displayed.
With the second type, the horizontal divergence hdiv of the wind, the vorticity hcurl of the wind
can be displayed with or without the streamlines of wind.
The four last buttons are aimed at controlling the type of view that can be obtained from GrADS.
With the rst the H button, one obtains a latlon view of the area selected see below for the
dierent area.
With the N button, one obtains a two dimension graph with the longitude on the x-axis and the
pressure level on the y-axis.
With the E button, one obtains the same two dimension graph but with the latitude on the
x-axis.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
15
With the P button, one obtains a graph with time on the x-axis and pressure level and y-axis for
a given point of the map. For the last three buttons N,E or P, after clicking on one of them,
the program will ask the user to click on a point of the map.
Time and Pressure level: This button bar right side of the screen controls the pressure level
and the time for which the data are plotted. It is enabled only in the latlon view H mode, since
there is no interest of having these two buttons bar in any of the three other display modes.
Controls: The top button bar should look like the one presented on Figure 4. The rst button
Figure 4: The top button bar
quit" was explained above. The next two buttons are used to zoom Z in or un-zoom Z a. When
Z in is clicked, the program asks in which region the user wants to zoom. Then one should click
in one of the two rectangles to see the zoomed area. The button Z a is then used to return to the
general map.
After that are the two menus controlling the display, GXout and MPJ. GXout handles the way data are
presented contours, shaded surfaces, vectors, streamlines,... and MPJ controls the type of projection
of the base map latlon, SPS, NPS or Robinson.
The 'wait' button is helpful if the user wants to print a hardcopy of the graphic windows. When
enabled, GrADS draws the map and the data normally, but waits for a user click before displaying
the buttons and the menus. It is generally used with the button 'print'. When enabled, this button
will make GrADS produce a meta-
le each time a new data is plotted. The meta-
les produced have
generic names: mfxxx.meta, where xxx stand for the number of the meta
le during this session of the
application. The user must be aware that GrADS will print the whole graphic screen, including the
menus and the buttons if 'wait' is not enabled.
The Zlog" button is useful only in the sections displaying modes N, E and P. It draws the y-axis
the pressure level, with a logarithmic scale.
The SMOT" button allows GrADS to perform a smoothing between the points of the grid, to smooth
the display.
The GRID" button draws on the map the grid on which the computation were made.
The 'ANIM' button cycles between the time steps to produce an animation indexed by time only in
H, N and E modes.
2.4
Background applications
Some applications have been developped to be executed in background, in order to produce images
and text les for the Web pages. These applications are described more precisely in the the part II.
Basically, there are rst three scripts designed to produce all the images shown on the Web pages.
Those applications, whose names are script1.gs, script2.gs and script3.gs, produce the latlon" maps
for the Web pages. The rst one deals with the general south paci
c region, the second deals with
16
Meteorological Prediction Software: User Manual
the zoomed region above Paranal and Chajnantor, and produces the reconstructed radio proles for
those two sites, and the third deals with the site above La Silla, and produces the reconstructed radio
proles above this site. Those applications are called by Unix scripts to be executed automatically
once per day to update the Web site. Normally, the user should not have to execute these scripts
himself. If the update of the Web pages was not made correctly, one should look for the les scrWeb0
and scrWeb4 and execute them instead of executing directly the GrADS applications.
Another set of background applications contains the applications designed to produce the text les
for the forecast verication pages. The rst one, verif0.gs, is called by an Unix-script every day to
produce the data for the graph and verication pages. The second one, writetxt.gs, produces the
kalman-corrected le, for the java applet presented on the verication pages. Again, the user should
not have to call those applications directly. The main script for the verication is called step2 and
is executed once per day. It updates the les te.log, pr.log and ws.log, which are the log les for the
temperature, the pressure and the wind speed. If there was a malfunction in those scripts, one should
know that there are some backup of those three critical les: the names of the backup are te.log.bak,
pr.log.bak and ws.log.bak.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
17
3 Satellite products
3.1
Introduction
In order to display a complete set of tools for the weather forecasts for the three site of interest, ESO
has an external contract with an american institue, CIRA 6. This institue provides ESO with two
types of images, one taken at the wavelength of 10.7 m, and which enables to visualize the cloudiness,
and the other is taken at 6.7 m, and enables to measure the precipitable water vapor present in the
atmosphere PWV and the amount or cirrhus clouds as well.
The second type of images is the object of an external contract with Andre Erasmus [email protected] .
By merging the images with the predictions of wind from ECMWF, it produces a forecast of which
pixel will be above the sites of La Silla and Paranal in the next hours. By converting its gray level
into PWV, one gets the forecasts of PWV. With the rst type of images, we just extract some zone
of interest and publish it on the Web as it will be explained in the next section.
3.2
The McIdas IDL front end utility
3.2.1 The execution of this utility
The images, provided by the CIRA institute, are coded with the McIdas format. In fact, it contains
more information than in a simple image. As the images are a view from earth taken at 36.000 km,
the projection is not a linear one so each point contains, in addition of its gray level, the coordinates
it represents on the Earth. One can not just take an image le" and try to open it with an image
browser. One needs an utility, just like GView, the one provided by CIRA and which runs under IDL7.
To start this utility, one needs rst to start the IDL software on olasg in any directory. The path is in
the environement variables, so one just need to type 'idlde' at the command prompt to start IDL. A
Graphical interface will appear, with a bottom line used as a command line. So in this one, one just
type 'gview' and IDL reads the program les, compile it and then execute the GView Utility.
3.2.2 The manual use of the GView utility
The use of the Gview utilty is pretty simple. Nevertheless, one can access an online documentation
of the software in the HTML format. The head le is GVIEW.HTM, and is located in the directory
diskbasmprediCIRAsoftwarehtml. Meanwhile, an abstract is presented here to help the user in
the main functions of the utility.
When the utility starts, one should see a menu similar to the one shown on gure 5. From there, one
can open a McIdas le, save the image in numerous formats Raw, Ti, Jpeg, Jpeg 24 bits, PGM,
PPM, Postscript, Color Table button File", manipulate the color tables button Color", load some
graphic les to add to the image button Graphic", diplay the image loaded enabled only after the
reading of a McIdas le button Display", display only part of the image button Sector", and
browse through the dierent images present on the disk buttons Prev" and Next".
Cooperative Institute for Research in the Atmosphere, Colorado State Universtity, Fort Collins CO 80523, USA,
URL: http:www.cira.colostate.edu
7
Interactive Data Language, Research Systems, URL:http:www.rsinc.com
6
18
Meteorological Prediction Software: User Manual
Figure 5: The Gview main button bar
Figure 6: The Gview display button bar
After the loading of an image and its display with the button Display", one can see the button bar
presented on gure 6. This button bar allows the user to apply some treatments to the image le.
Those special eects are described in the online documentation of the products. One just needs to
know that it enables to user to invert the colors button Invert", to manipulate the color histogram
button Equalize" and to apply some enhancements. Meanwhile, in this version provided by CIRA,
some buttons are ineective, like the Resize" button, the resize function not being yet implemented.
3.3
Bakground applications
Some applications were developped to go to the site of CIRA, download the images and apply some
treatments on them for later publishing on the Web. Those applications consist in one script le on
olasg, satellite and one main IDL application, tryRead.pro, which uses a startup IDL le satellite.pro,
designed to set some global variables.
The experience shows that the sensitive point in those operations is to take the images from the
CIRA ftp site to the ESO 'olasg' machine. The link being sometimes very slow, the 'ftp' sometimes
cuts down the transmission, and the le is not transmitted properly, which results in a hole" in the
satellite image sequence. A log" le is located in the directory diskbasmprediVerif, under the
name satellite.report, to present the correct transmission and the failures.
The other parts of the process, like the processing of the images and their publishing on the web pages,
seems not to pose any problem. The path followed to process the images is then the following: satellite
unix script which goes to get the images on the CIRA site, satellite.pro IDL applications, which
sets global variables and calls TryRead.pro IDL application which processes the image. Those three
step are located on olasg. The Unix script located on the web4 machine is used to get the processed
images on olasg and publish them on the web pages.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
19
4 The Local products
4.1
Introduction
To verify the forecast, one needs to access some measured data to compare with the forecasts. ESO has
installed meteo-monitors on top of the sites of interest for now La Silla and Paranal. Those meteo
monitors take mesures every minute and update a database containing meteorological data. This database is located on the olasg machine at ESO, and maintained by Benoit Pirenne [email protected]
The goal for us is to access those real data for times matching times of predictions by ECMWF. In
doing so we have for each time of prediction" one triplet of points: the forecasts of ECMWF, the
analysis of ECMWF run of their model for time T=0 and the measured data 20 minutes average.
4.2
Background applications
To insure the update of the les containing the triplets of points as seen before, some Unix scripts are
used to get the local data, to extract the forecasts and the analysis and to run the Kalman lter on
the predictions. The rst script, which goes in the meteo database and get the data for time matching
those of ECMWF 00 UT,06 UT,12 UT,18 UT, is located on the olasg machine in the directory
diskb asmpredi Verif LaSilla or
diskb asmpredi Verif Paranal for now the data for Paranal is not available, but the scripts have
been readied, to insure the correct functionning as soon as data become available. This script, named
getlastmeteo, gets the measured data of the ground temperature, the wind speed at ground level and
the ground pressure, and record them in 'log' les nammed te.log, ws.log and pre.log. Those les are
located in the directories mentionned above.
The structure of the les is quite easy to understand. For each record each line, it contains the
date, the hour, the measured data, the analysis of ECMWF, the closest forecast from ECMWF and
the kalman corrected forecast see later for explanations concerning the kalman lter. Those log les
are maintained by a set of scripts: step1, which calls the getlastmeteo script, step2, which lls the
analysis and forecasts column and calls step3, which lls the kalman corrected column by calling a
GrADS application: verif0.gs. The latter calls the two user dened functions cuspLaSi or cuspPara,
for LaSilla and Paranal. The second step, step2, calls also another GrADS application, writetxt.gs,
which deals with the production of a text le needed by the verication java applet. Then it calls at
the end a gnuplot application to produce some graphics presented on the verication pages step4.gp .
20
Meteorological Prediction Software: User Manual
5 The internet pages
5.1
Overview
In order to suit the needs of the astronomer, the data sent by ECMWF had to be postprocessed and
displayed graphically. This has been done on a web site dedicated to those data and automatically
updated twice a day with new images and graphs.
The URL of the web site is :
http:www.eso.orggen-facpubsastclimforecastmeteo
Browsing the site is quite intuitive, since all is well explained and all the links are well dened.
On the site, the user will be able to see the predictions for the chilean coast, for the regions of Paranal
and La Silla, both for short and long term. The user will nd some time indexed graphics, to follow
and anticipate the evolution of the situation above the three sites of Paranal, La Silla and Chajnantor.
On the rst page, the user can choose to see some more quantitatives pages dealing with what is called
the corrected forecasts and the forecast verication. For the rst topic, the corrected forecasts, the
basic reason is that ECMWF uses a grid for their predictions which is too coarse. It means for example
that for them the site of Paranal is located on the sea they use a 0.5 degree grid, which corresponds
to 60 km
. So we apply a kalman lter, which analyses the dierence between the forecasts and the
measures over the past 14 days, and corrects in an appropriate way the next forecasts. The results
are in agreement with the ones CRS4 obtained during the feasability study CRS4 97.
For the second topic, the forecasts verication pages contain statistics about the accuracy of the raw
predictions, compared to the accuracy of the kalman-corrected predictions.
5.2
5.2.1
Web publishing scripts
ECMWF products
The update of the site is made automatically by dedicated scripts which are described in the following
pages. These scripts are on the olasg machine, in the directory home2asmprediECMWFdatatmp.
The top level scripts are named scrWeb0 and scrWeb4, each one taking care of one of the time period
0 for the 00Z predictions and 4 for the 12Z ones
.
Basically, these scripts look for the last set of les received from ECMWF. Then they process those
les with a GrADS application in order to produce the images which will be published on the web.
These images, rst in a metale format, are processed to produce .gif les which are then transferred
to the web4 machine through a ftp canal by scripts named publish0 and publish4 located on the web4
machine.
The GrADS applications used are script1.gs, script2.gs and script3.gs, which are described in this
document see section 2.4
.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
21
5.2.2 CIRA products
As seen in the introduction, CIRA is providing satellite images to ESO in order to follow the evolution
of the cloud cover on the south pacic region and on the chilean coast. Those images are processed
to produce eight images which are published on the Web. On the rst page of the satellite images,
Figure 7: The Satelitte applet
the user can view the last updated image. By clicking on the link just above the image, the user has
access to a java applet, which lets him see the last eight images 24 hours
. The use of the applet is
really simple see g.7, the user can browse between the images buttons next and previous
. He can
also start an animation of the eight images the time between each frame can be regulated by the text
box below the rst button bar
. He can stop the animation and reset the images it displays the rst
image and resets the image counter
.
Another thing the user can do is to click on the button zoom to display a more accurate view of the
Chilean coast in the area of the observatories.
5.2.3 Verication pages
The verication pages are provided to present the results of the corrected forecasts. As said in the
introduction, the corrections are made by a kalman-lter, initiated by 14 days of predictions, local
measures and already corrected forecasts. The program is described in the users manual see sec. 6
.
The navigation on the verication pages is really easy, the rst pages are made with three graphs
pressure, temperature and wind speed
to show the evolution of the three type of meteorological data
forecasts, kalman corrected forecasts and measures
. There are two versions of this rst page. On
the rst is added a small java applet to show the starting and ending dates of the graphs, and then to
show the next kalman corrected forecasts. The user can choose the date, the hour of prediction and
the type of parameter to display.
On the second page are presented three more graphs showing the matches with real data for the three
22
Meteorological Prediction Software: User Manual
parameters and the accuracy of the kalman corrected forecasts with respect to the raw forecasts. These
pages should provide the user with some informations about the accuracy of the predictions during
the last 15 days. It might be useful to determine if the predictions are reliable or not on the long term,
therefore a verication archive has been provided in the directory diskb asmpredi Verif LaSilla on
the olasg machine.
23
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
6 Fortran programs used to produce the forecasts
To improve the forecasts quality for a better accuracy with the real measured data, some programs
were developped. Those programs were written in FORTRAN to be consistent with the ECMWF
data processing software. In fact, three programs were developped.
The rst set of GrADS user dened functions biliLasi, biliPara and biliChaj is aimed at
rening the predictions. As the reader has seen before, the predictions are made for equidistant
points on a grid. When we ask for the prediction for a point which is not on the grid, GrADS
performs a round-up with the nearest point on the grid. To improve that, CRS4 proposed to
use a bilinear interpolation. That is to say that when the user requests one point, GrADS
takes the four surrounding points and performs an interpolation with respect to the distance
between the requested point and the grid points. This type of interpolation is known as a bilinear
interpolation and can be described with the following formulas: Given the following stencil,
F 4x1 y 2F 3x2 y2
F x y F 1x1 y 1F 2x2 y1
With F1, F2, F3 and F4 being the forecasts made by ECMWF at the 4 nearest points and F
the desired value, we use the equation
F x y = 1 t 1 u F 1 + t 1 u F 2 + t u F 3 + 1 t u F 4
with
t = xx2 xx11 u = yy2 yy11
The second set of GrADS user dened functions cuspLaSi, cuspPara and cuspChaj is quite
similar to the rst one, except that it performs an interpolation on the altitude to get the correct
forecast for the given altitude. The interpolation retained for the altitude is a cubic spline one,
which seems to give better result than the bilinear one in this case. Those functions also write
down the 'log' les with those exact results. Then it should never be used by the user, because
a bad use couls corrupt the log les for the three parameters temperature, wind speed and
pressure.
One needs to know that, as the ECMWF may be sometimes corrupted or may contain some
errors, there is a protection in this interpolation. That means that if the interpolated data is out
of ranges dened after, it will be markes as unavailable 999. in the log les. Moreover, as the
pressure is a key for the interpolation for the temperature and the wind speed, if the pressure
value is out of ranges, the three data are marked as unavailable. The ranges are:
0 windspeedm=s 99
99 pressuremB 1099
99 temperature C: 99
The cubic spline interpolation is based on the following formulas more accurate explanations
may be found in the Numerical Recipes Press et al. 94. Suppose that f p is the vertical
prole of ECMWF data. We compute f p, with p 2 p p +1:
o
i
j
j
f = A p + B p +1 + C p + D p +1
j
j
00
00
j
j
24
Meteorological Prediction Software: User Manual
with
A = ppj+1 pp B = p p pjp
j +1
j
j +1
j
C = 61 A3 Apj+1 pj 2 D = 16 B3 Bpj+1 pj 2
2
p = ddpf2
00
Figure 8: Principle of the Kalman lter
The third program Kalman was developped upon a basis provided by CRS4. It performs a
Kalman correction. The Kalman lter is a non linear learning lter, which means that basically it
computes the dierence between the forecasts or other parameters see below with the measured
data, and then corrects the next forecasts assuming a similar trend will be observed. For the
current problem, CRS4 showed, in their feasibility study CRS4 97, that the best results were
obtained by taking into account the forecast from ECMWF, the analysis which are the run
of the model for T = 0, the measured data and the last corrected forecasts. The principle of
this lter is shown on the Figure 8. The initialization les for this lter are the three '.log' les
te.log, pr.log and ws.log for temperature, pressure and wind.
A manual mode is available on olasg. The user should go to directory diskbasmprediVerifLaSilla and type kalman. The user should then provide on the command line the type of data to correct
e.g. kalman ws, kalman te or kalman pr , and then it asks the user for the input value to
correct.
In a background mode, this program is called by the script step2 to produce the statistics shown
on the Web pages.
25
Part II
Maintenance manual
7 Maintenance of the scripts
7.1
Scripts related to ECMWF products
7.1.1 The transfer scripts ECMtoOlasg and ECMtoOlasg2
Those two scripts are on the opus3 machine, in the account asmpredi. They are used to transfer the
ECMWF data les from the opus3 machine where ECMWF put them to the olasg machine where
we process and store them. In this section we will just comment on one of the scripts, the other being
quite similar.
The scripts execute each 20 minutes 00, 20 and 40 for ECMtoOlasg and 10, 30 and 50 for ECMtoOlasg2 .
The crontab commant is the following :
00,20,40 * * * * homeasmecmwfECMtoOlasg devnull 2&1
10,30,50 * * * * homeasmecmwfECMtoOlasg2 devnull 2&1
The listing of the script follows:
EcmtoOlasg
#!bintcsh
# running on opus3, this script uploads the prediction files
# from ECMWF
# and put them on olasg in
# home2asmprediECMWFdatadata00UT for
# the 00ut predictions
# and in home2asmprediECMWFdatadata12UT for the 12UT
# predictions.
# On opus3 it creates an archive with the files uploaded and
# deletes those ones.
cd homeasmecmwf
# goes on olasg to see the last day of transfer
ftp -n olasg EOD essai
user asmpredi xxxxxxx
bin
prompt
cd home2asmprediECMWFdatadata12UT
ls -al
quit
EOD
#
grep ESD36 essai essai2
rm essai
# we search the last day of transfer the last file
# in the list
@ d = `tail -5c essai2`
set dd = `expr $d - 4`
set d = `expr $dd 10`
rm essai2
# we must transfer all the files for the completed
# days since this date
26
Meteorological Prediction Software: Maintenance Manual
set eend = 0
while $eend != 1
# we add 1 to d
set d = `expr $d + 1`
if $d 10 set d = 0$d
if $d 100 set d = 0$d
# if d = 367 then d = 001
if $d == 367 set d = 001
# is there a complete block for the d-day ?
@ f = `ls -al ESD36$d4 | wc -l`
if $f != 0 then
ftp -n olasg EOD
user asmpredi xxxxxxx
bin
hash
cd home2asmprediECMWFdatadata12UT
mput "ESD??$d4"
quit
EOD
tar c ESD??$d4 ESD$d4.tar
gzip ESD$d4.tar
ftp -n olasg EOD
user asmpredi xxxxxxx
bin
hash
cd diskbasmprediECMWFarchives12UT
mput "ESD$d4.tar.gz"
quit
EOD
rm -f ESD??$d4
rm -f ESD$d4.tar.gz
else
set eend = 1
endif
end
Those scripts are quite simple : ECMWF transmits the les in a sequence. Once the le ESD20???0
for the 00Z products and ESD36???0 for the 12Z products are there, we have a complete set of
les which we can transmit to olasg.
On olasg we put the les in the directory
home2asmprediECMWFdatadata00UT or data12UT. After doing that we create an archive with
the complete leset, we 'gzip' it and we put it on olasg in diskbasmprediECMWFarchives00UT
or
diskbasmprediECMWFarchives12UT. In fact the script loops until it has treated the complete
leset, incrementating on the number of the day.
7.1.2 The display scripts scrWeb0 and scrWeb4
Those two scripts one for each type of product are located on the olasg machine, in the directory
home2asmprediECMWFdatatmp. They are used to process the ECMWF les located on olasg
to produce the images needed by the Web pages.
Those scripts are executed every 30 minutes 0 and 30 for scrWeb0 and 15 and 45 for scrWeb4 . The
crontab command is:
00,30 * * * * home2asmprediECMWFdatatmpscrWeb0 devnull 2&1
15,45 * * * * home2asmprediECMWFdatatmpscrWeb4 devnull 2&1
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
The listing of the scrWeb0 script follows :
scrWeb0
#!bintcsh
#
# Script designed to produce gif files to
# update web page on forecast.
#
set echo
cd home2asmprediECMWFdatatmp
#
ls -al home2asmprediECMWFdatadata00UT essai
grep ESD20 essai essai2
rm essai
# we search the last day of transfer the last file in the list
@ d = `tail -5c essai2`
set d = `expr $d 10`
rm essai2
# we loook for the last day of the update
set lastd = `more day0.txt`
if $lastd $d || $lastd == 366 || $lastd == 365
&& $d == 1 then
if $d 10 set d = 0$d
if $d 100 set d = 0$d
cp home2asmprediECMWFdatadata00UTESD??$d0 .
ECMextrDAT $d 0
rm ESD??$d0
grads -lb -c 'run script1.gs '$d' 0 0'
foreach file *.meta
gxgif -r -i $file
end
rm *.meta
#
grads -lb -c 'run script2.gs '$d' 0 0'
foreach file *.meta
gxgif -r -i $file
end
rm *.meta
#
grads -lb -c 'run script3.gs '$d' 0 0'
foreach file *.meta
gxgif -r -i $file
end
rm -f *.meta
rm -f ESD$d0_*
rm day0.txt
cat day0.txt EOD
$d
EOD
set date = `date +Cy.m.d.H:M`
awk 'print "ECMWF.result" END printf
"s : Successfull traitment of day d, product of 00UTn",
"'$date'","'$d'" "ECMWF.result"' ECMWF.result
endif
set currd = `date +j`
if $lastd $currd - 1 then
set date = `date +Cy.m.d.H:M`
set lastd = `expr $lastd + 1`
awk 'print "ECMWF.result" END printf
"s : There has been an error, no traitment for day d, 00UTn","'$date'",
"'$lastd + 1'" "ECMWF.result"' ECMWF.result
rm day0.txt
cat day0.txt EOD
$lastd
EOD
endif
#
27
28
Meteorological Prediction Software: Maintenance Manual
cp ECMWF.result diskbasmprediVerif
The scrWeb4 script is quite similar to the rst one. It looks in a le day0.txt or day4.txt for the
12Z product for the last day which has been processed. If the current day is superior and if there
is a complete set of les for this day, it processes the set of les : it uses the utility ECMextrDAT
to extract the information from the les, and then calls the GrADS utility with the name of scripts
designed to perform meteorological treatments. Thoses scripts produce '.meta' les which we process
with the utility gxgif designed to convert meta les to gif les. Then we can remove all the temporary
les, and write in a le ECMWF.result if there has been an error or not.
7.1.3 The verication scripts step1, step2, step3, getlastmeteo, step4.gp
These are scripts dedicated to the forecasts verication. They are on the olasg machine, in the directory
diskbasmprediVerif and their aim is to produce some data les for gnuplot. This utility is used to
draw the plots used in the Web pages, under the link 'forecasts verication'.
The rst script, step1, is used to call an utility, getlastmeteo designed to go into the archeso database
and collect the corresponding real measured data to compare with the forecasts of ECMWF. The
listing of getlastmeteo follows :
getlastmeteo
#! bintcsh -f
# modified by M.Albrecht 191098
cd diskbasmprediVerif
if $1 == "" then
echo "Usage: getlastmeteo db-password hour site date"
echo "If no date is supplied, today is assumed"
exit
endif
set hour = $2
set pswd = $1
set site = $3
if $site != "lasilla" && $site != "paranal" then
echo "Wrong site: must be either 'lasilla' or 'paranal'"
echo "Usage: getlastmeteo db-password hour site date"
exit
endif
if $4 == "" then
set date = `date -u +Cy.m.d`
else
set date = $4
endif
set
set
set
set
datehour = $date.$hour
hoursec=`~flowmgrbincvdate -$datehour `
hourstart = `expr $hoursec - 600`
hourend = `expr $hoursec + 600`
isql -Uwww -P$pswd EOF temp
use ambient
go
select avgt310. from meteo_$site
where start_date between -$hourend and -$hourstart and t3 != -32768
select avgpr10. from meteo_$site
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
29
where start_date between -$hourend and -$hourstart
and pr != -32768
select avgws310. from meteo_$site
where start_date between -$hourend and -$hourstart
and ws3 != -32768
go
EOF
# If seeing is also wanted uncomment the lines below
#isql -Uwww -P$pswd EOF temp
#use ambient
#go
#select avgfwhm100. from seeing_$site
# where start_date between -$hourend and -$hourstart
#EOF
set t1 = `awk 'NR == 3 print' temp`
if $t1 != "NULL" then
awk 'print "te.log" END if '$hour' 10 printf"s
0d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour','$t1',999,999,999 "te.log" else printf"s
d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour','$t1',999,999,999 "te.log"' te.log
else
awk 'print "te.log" END if '$hour' 10 printf"s
0d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour',999,999,999,999 "te.log" else printf"s
d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour',999,999,999,999 "te.log"' te.log
endif
set pr = `awk 'NR == 8 print' temp`
if $pr != "NULL" then
awk 'print "pr.log" END if '$hour' 10 printf"s
0d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour','$pr',999,999,999 "pr.log" else printf"s
d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour','$pr',999,999,999 "pr.log"' pr.log
else
awk 'print "pr.log" END if '$hour' 10 printf"s
0d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour',999,999,999,999 "pr.log" else printf"s
d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour',999,999,999,999 "pr.log"' pr.log
endif
set ws = `awk 'NR == 13 print' temp`
if $ws != "NULL" then
awk 'print "ws.log" END if '$hour' 10 printf"s
0d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour','$ws',999,999,999 "ws.log" else printf"s
d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour','$ws',999,999,999 "ws.log"' ws.log
else
awk 'print "ws.log" END if '$hour' 10 printf"s
0d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour',999,999,999,999 "ws.log" else printf"s
d
6.1f
6.1f
6.1f
6.1fn","'$date'",'$hour',999,999,999,999 "ws.log"' ws.log
endif
rm temp
It goes into the archeso database to extract the data corresponding to the 00UT, 06UT, 12UT and
18UT measurments of the day and append them in the three log les pr.log, te.log, ws.log and their
backup
containing the data. In a rst step, only the rst row of the log le is lled with the measured
data, the others are lled with '999.'.
The script getlastmeteo is called by another script, step1, for which the listing follows :
step1
#!bintcsh
# This shell script performs the first step in building
# the data file for Kalman-Verif.
# It uses the script getlastmeteo to build the files
# pr.log, temp.log and ws.log
#
30
Meteorological Prediction Software: Maintenance Manual
cd diskbasmprediVerif
#
#mv te.log temp.te
#tail -n 100 temp.te te.log
#mv pr.log temp.pr
#tail -n 100 temp.pr pr.log
#mv ws.log temp.ws
#tail -n 100 temp.ws ws.log
#
if $1 == "-help" || $1 == "-h" then
echo "Usage: step1 date"
echo "If no date is supplied, today is assumed"
echo "Date must be written yyyy.mm.dd"
exit
endif
#
set date = $1
getlastmeteo ####### 0 lasilla $date
getlastmeteo ####### 6 lasilla $date
getlastmeteo ####### 12 lasilla $date
getlastmeteo ####### 18 lasilla $date
The script step1 is located on the olasg machine, in the directory diskbasmprediVerif. It is executed
once per day, at 21h local time MET DST. The crontab line follows:
00 21 * * * diskbasmprediVerifLaSillastep1 devnull 2&1
One must notice that the three log les are not limited in size to allow for long term drifts detection.
After creating the log les with the step1 script, we need to complete them with the analysis and the
predictions by ECMWF. Those jobs are done by a second script, step2, which complete the log les.
The listing of this script follows :
step2
#!bintcsh
#
# This script will add the second part of the .log files,
# i.e. the analysis and the predictions of ECMWF.
# It uses the user defined function cuspLaSi for La Silla
# to extract "exact" analysis and predictions for the site
# It performs a bilinear interpolation on lat lon and a
# cubic spline interpolation on the altitude.
#
cd diskbasmprediVerif
#
# This part looks for the last set of files from ECMWF
#
ls -al home2asmprediECMWFdatadata00UT essai
grep ESD20 essai essai2
rm essai
# we search the last day of transfer the last file in the list
@ d = `tail -5c essai2`
set d = `expr $d 10`
rm essai2
if $d 10 set d = 0$d
if $d 100 set d = 0$d
cp home2asmprediECMWFdatadata00UTESD??$d0 .
ECMextrDAT $d 0
rm ESD??$d0
#
# We look for the day before in oder to be able to add the predictions
#
set echo
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
set dbef = `expr $d - 1`
set year = `date -u +Cy`
if $dbef == 0 then
if `expr $year 4` == 0 && `expr $year 100` != 0 || `expr $year 1000` == 0 then
set dbef = 366
else
set dbef = 365
endif
endif
if $dbef 10 set dbef = 0$dbef
if $dbef 100 set dbef = 0$dbef
cp home2asmprediECMWFdatadata00UTESD??$dbef0 .
ECMextrDAT $dbef 0
rm ESD??$dbef0
#
# Now we have the files, we can process them to produce the log files
#
grads -lb -c 'run verif0.gs '$d' 0'
grads -lb -c 'run verif0.gs '$dbef' 0'
rm ESD*
#
# Second passage for the predictions of 12 hour
#
ls -al home2asmprediECMWFdatadata12UT essai
grep ESD36 essai essai2
rm essai
# we search the last day of transfer the last file in the list
@ d = `tail -5c essai2`
set dd = `expr $d - 4`
set d = `expr $dd 10`
rm essai2
if $d 10 set d = 0$d
if $d 100 set d = 0$d
cp home2asmprediECMWFdatadata12UTESD??$d4 .
ECMextrDAT $d 4
rm ESD??$d4
#
# We look for the day before in oder to be able to add the predictions
#
set echo
set dbef = `expr $d - 1`
set year = `date -u +Cy`
if $dbef == 0 then
if `expr $year 4` == 0 && `expr $year 100` != 0 || `expr $year 1000` == 0 then
set dbef = 366
else
set dbef = 365
endif
endif
if $dbef 10 set dbef = 0$dbef
if $dbef 100 set dbef = 0$dbef
cp home2asmprediECMWFdatadata12UTESD??$dbef4 .
ECMextrDAT $dbef 4
rm ESD??$dbef4
#
# Now we have the files, we can process them to produce the log files
#
grads -lb -c 'run verif0.gs '$d' 4'
grads -lb -c 'run verif0.gs '$dbef' 4'
cp te.log te.log.bak
cp pr.log pr.log.bak
cp ws.log ws.log.bak
#
# Call for the third part of the script
#
diskbasmprediVerifstep3
#
# Third party script to create the data file for the java pages
#
rm data.txt
31
32
Meteorological Prediction Software: Maintenance Manual
grads -lb -c 'run writetxt.gs '$d' 4'
diskbasmprediVerifdata
#
rm ESD*
home2sofitoolsbingnuplot .step4.gp
In step2, a call is made with the current and the previous day to ll the column of the analysis of the
current day and of predictions of the day before for the current day. step2 calls the ECMextrDAT
utility with the correct sets of les, and processes them through verif0.gs, to extract the correct
information.
The third step is to add the kalman corrected forecasts in the three log les. This is the task of the
script step3. The listing follows :
step3
#!binsh
#
# Part 3 of the verification script, designed to
# compute the kalman results for the verification
# phase.
#
# First part : file te.log
#
while read -r date hour real ana predi kal
do
if $kal -eq 999. && $predi -ne 999. then
cat EOD temp
$predi
EOD
`kalman te temp temp2`
kal=`tail -8c temp2`
rm temp
rm temp2
fi
if $hour -lt 10 then
printf "s
0d
6.1f
6.1f
6.1f
else
printf "s
d
6.1f
6.1f
6.1f
fi
done te.log temp3
mv temp3 te.log
#
# Second part : file pr.log
#
while read -r date hour real ana predi kal
do
if $kal -eq 999. && $predi -ne 999. then
cat EOD temp
$predi
EOD
`kalman pr temp temp2`
kal=`tail -8c temp2`
rm temp
rm temp2
fi
if $hour -lt 10 then
printf "s
0d
6.1f
6.1f
6.1f
else
printf "s
d
6.1f
6.1f
6.1f
fi
6.1fn" $date $hour $real $ana $predi $kal
6.1fn" $date $hour $real $ana $predi $kal
6.1fn" $date $hour $real $ana $predi $kal
6.1fn" $date $hour $real $ana $predi $kal
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
done pr.log temp3
mv temp3 pr.log
#
# Third part : file ws.log
#
while read -r date hour real ana predi kal
do
if $kal -eq 999. && $predi -ne 999. then
cat EOD temp
$predi
EOD
`kalman ws temp temp2`
kal=`tail -8c temp2`
rm temp
rm temp2
fi
if $hour -lt 10 then
printf "s
0d
6.1f
6.1f
6.1f
else
printf "s
d
6.1f
6.1f
6.1f
fi
done ws.log temp3
mv temp3 ws.log
33
6.1fn" $date $hour $real $ana $predi $kal
6.1fn" $date $hour $real $ana $predi $kal
kalman utility to correct the forecast. It reads the second data column of the
log le Pr.log, te.log or ws.log corresponding to the type of data to correct and extracts the value
of the forecast, then writes the Kalman corrected value in the last column. It does this for the three
parameters, hence the three parts in the script.
This script calls the
7.1.4 The publication scripts publish0,publish4
To end the tour of the ECMWF scripts, we must cite the publication scripts, located on the web4
machine, which download the image les from olasg, merge them to produce some 'animated gifs' and
then copy those les on the web machine web1 and web2 with the utility webcp. There are two
scripts on web4, publish0 and publish4 for the 00UT and the 12UT products.
We print here only one of the two scripts, the other is quite similar to this one.
publish0
#!bintcsh
# this script is designed to produce animated gifs from gif files
# and to publish both of them on the web for the 12UT predictions
cd disksdiskwawebdocsesonewgen-facpubsastclim
cd .forecastmeteoECMWFshortimages
#
set echo
ftp -n olasg.hq.eso.org EOD
user asmpredi xxxxxxx
bin
prompt
cd home2asmprediECMWFdatatmp
mget temp?.gif
mdelete temp?.gif
mget temp?_zoom1.gif
mdelete temp?_zoom1.gif
mget temp?_zoom2.gif
mdelete temp?_zoom2.gif
mget gh?.gif
34
Meteorological Prediction Software: Maintenance Manual
mdelete gh?.gif
mget gh?_zoom1.gif
mdelete gh?_zoom1.gif
mget gh?_zoom2.gif
mdelete gh?_zoom2.gif
mget wind?.gif
mdelete wind?.gif
mget wind?_zoom1.gif
mdelete wind?_zoom1.gif
mget wind?_zoom2.gif
mdelete wind?_zoom2.gif
mget *_modwin.gif
mdelete *_modwin.gif
mget *_vecwin.gif
mdelete *_vecwin.gif
mget *_temp.gif
mdelete *_temp.gif
mget *_gh.gif
mdelete *_gh.gif
quit
EOD
#
rm temp.gif temp_zoom1.gif temp_zoom2.gif gh.gif gh_zoom1.gif gh_zoom2.gif wind.gif wind_zoom1.gif wind_zoom2.gif
......scriptsgifmerge temp?.gif temp.gif
......scriptsgifmerge temp?_zoom1.gif temp_zoom1.gif
......scriptsgifmerge temp?_zoom2.gif temp_zoom2.gif
......scriptsgifmerge gh?.gif gh.gif
......scriptsgifmerge gh?_zoom1.gif gh_zoom1.gif
......scriptsgifmerge gh?_zoom2.gif gh_zoom2.gif
......scriptsgifmerge wind?.gif wind.gif
......scriptsgifmerge wind?_zoom1.gif wind_zoom1.gif
......scriptsgifmerge wind?_zoom2.gif wind_zoom2.gif
usrserverbinwebcp *.gif
Another script used for the publication is the verif script located on the web4 machine. It is used
to download all the images and data les lasi.txt, and later para.txt and chaj.txt to the Web pages.
The listing is not included because this script only performs ftp.
7.2 Scripts related to CIRA products
7.2.1 The satellite scripts satellite, goesGetSat
The satellite script is located on the olasg machine, in the directory
home2asmprediCIRAutils. It goes on tanager a CIRA machine - one must notice that the
password has been disabled in the listing ! and looks for the last image. If there is on tanager a more
recent image than the one we have on olasg, it downloads it and performs some treatments on it with
the help of the IDL utility. The IDL scripts produces two images, clouds.gif and zoom.gif. The aim
of the web pages being to show eight images of each type, we create a sequence by shifting at each
execution the sequence number of the image.
The listing follows :
satellite
#!bintcsh
#
# This script is designed to get the last image from
# the goes8 satellite and to publish it on the web
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
#
cd home2asmprediCIRAutils
#
# We check if there is a new file to transfert
#
cd home2asmprediCIRAutils
# we check if there is currently a transfer running
set currFile = `tail -n +1 current`
if $currFile != "" goto end
# No transfer running
# We check if there is a new file to transfert
#
rm temp
ftp -n tanager.cira.colostate.edu EOD temp
user eso goes8
cd andre
bin
prompt
ls
quit
EOD
#
# we look for the last .c04 file
#
rm temp2
grep "c04" temp temp2
set temp = `tail -c 55 temp2`
rm tempString
cat tempString EOD
$temp
EOD
set size = `cut -c1-8 tempString`
set filegz = `tail -c 26 tempString`
set lastFile = `tail -n +1 last`
set currFile = `tail -n +1 current`
#
if $filegz != $lastFile && $filegz != $currFile then
#
rm current
cat current EOD
$filegz
EOD
set file = `cut -c1-22 current`
#
# we go and get the last file
#
ftp -n tanager.cira.colostate.edu EOD
user eso goes8
cd andre
bin
prompt
get $filegz
quit
EOD
#
# we check if the file was transfered correctly sizenew = sizeold?
#
rm temp
ls -al $filegz temp
set sizeNew = `cut -c36-44 temp`
if $size == $sizeNew then
#
vltgnubingunzip $filegz
mv $file temp.c04
#
set date = `date '+Cy.m.d.H:M'`
awk 'print "satellite.report" END printf"s : succesfull loading of sn","'$date'","'$filegz'" "satellite.report"' satellite.report
rm last
cat last EOD
35
36
Meteorological Prediction Software: Maintenance Manual
$filegz
EOD
rm current
touch current
# now we run the idl script to transform this image
# in a gif file and to add coasts and points for
# the three sites
#
rm namefile
cat namefile EOD
$file
EOD
idl satellite.pro
#
mv clouds2.gif clouds1.gif
mv clouds3.gif clouds2.gif
mv clouds4.gif clouds3.gif
mv clouds5.gif clouds4.gif
mv clouds6.gif clouds5.gif
mv clouds7.gif clouds6.gif
mv clouds8.gif clouds7.gif
mv clouds.gif clouds8.gif
#
mv zoom2.gif zoom1.gif
mv zoom3.gif zoom2.gif
mv zoom4.gif zoom3.gif
mv zoom5.gif zoom4.gif
mv zoom6.gif zoom5.gif
mv zoom7.gif zoom6.gif
mv zoom8.gif zoom7.gif
mv zoom.gif zoom8.gif
#
set date = `date '+Cy.m.d.H:M'`
#
awk 'print "satellite.report" END printf"s : succesfull processing of sn","'$date'","'$filegz'" "satellite.report"' sa
else
#downloading was aborted by ftp timeout
rm $filegz
rm current
touch current
set date = `date '+Cy.m.d H:M'`
awk 'print "satellite.report" END printf"s : ftp abort of sn","'$date'","'$filegz'" "satellite.report"' satellite.rep
#
endif
else
# no new file posted on tanager
endif
#
#mv satellite.report satellite.temp
#tail -n 50 satellite.temp satellite.report
cp satellite.report diskbasmprediVerif
#
end:
Due to the unreliability of the internet link between ESO and CIRA, we had to implement some functions to verify, for example, that the size of the uploaded le was correct. Sometimes the transmission
is so slow that the ftp gets a timeout before transmitting the whole le. So we need to check if the
size of the uploaded le is correct. If not, we write a message in a log le and try again 10 minutes
later. This script runs every 10 minutes. If the le has not been successfully transfered before the
next image is posted on tanager, there is no recovery procedure currently implemented.
The log le satellite.report is located on the olasg machine, in the directory diskbasmprediVerif.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
37
7.2.2 The goesGetSat script
This script is located on the web4 machine, goes to olag to get the full size binned images and the
corresponding unbinned zoom images and publish them on the web.
The listing follows:
goesGetSat
#!bintcsh
#
# This script goes on olasg to get the satellite images and the
# corresponding animated gifs.
#
cd disksdiskwawebdocsesonewgen-facpubsastclimforecastmeteoCIRAimag
es
#
ftp -n olasg.hq.eso.org EOD
user asmpredi xxxxxxx
bin
cd home2asmprediCIRAutils
prompt
mget clouds*.gif
mget zoom*.gif
quit
EOD
#
cp clouds8.gif clouds.gif
usrserverbinwebcp clouds*.gif
usrserverbinwebcp zoom*.gif
38
Meteorological Prediction Software: Maintenance Manual
8 The GrADS applications and utilities
8.1 The front-end utility mainMeteo.gs
This utility is located on the olasg machine in the
home2 asmpredi ECMWF data tmp directory. As a nal product, it does not require a lot of
maintenance. It must be in the same directory as a copy of the ECMextrDAT utility. It must have
access to the data les from ECMWF, too. In fact at the startup of the script, it goes to the directory
.. data00UT or .. data12UT and copies the data les to the current directory. So if the utility is
moved, one must update the paths at the beginning of the script.
Another problem that may appear is an incompatibility with a future version of GrADS. Like all the
GrADS tools and scripts, it has been developped with the version 1.7 beta 7 and tested succesfully
with the version 1.7 beta 9. So if a new version of GrADS is installed , one must check the backward
compatibility with all the tools developped.
8.2 The 'images' scripts script*.gs
There are three scripts used to produce the meteorological predictions images for the web, one for
each region of interest. The rst, script1.gs deals with the complete area: the South Pacic and the
Chilean coast. The second, script2.gs, deals with the region above Paranal and Chajnantor. The
third, script3.gs, deals with the region above La Silla.
For each region, the script produces the lat-lon maps for the temperature at 700 mb, the wind streamlines at 200 mb and the geopotential height at 700 mb. The second and third scripts produce the
reconstructed radio proles above the sites of interest. Here follows the listing of the third script, the
rst is the same without the sites proles, and the second has one more site for the proles.
If one wants to produces new graphs for other levels, one just needs to change the 'set lev xxx' lines
for the new levels.
script3.gs
function mainargs
'reinit'
_day = subwrdargs,1
_hour = subwrdargs,2
_type = 0
if hour != 0
_type = subwrdargs,3
endif
file_ctl='ESD'_day''_hour'_'_type*3+3'.ctl'
'open 'file_ctl
'set display color white'
'set mpdset hires'
'c'
'set grads off'
'set clab forced'
'set parea 0.6 9.4 0.75 7.5'
'set gxout shaded'
'set csmooth on'
*** Drawing of the 7 or 17 images in the general region
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
if _hour = 0 | _hour = 4 & _type = 0
_nbFiles = 7
else
_nbFiles = 17
endif
'set lev 700'
i = 1
while i = _nbFiles
'c'
if _hour = 0 | _hour = 4 & _type = 0
'enable print temp'i'_zoom2.meta'
else
'enable print temp'i'bis_zoom2.meta'
endif
'set rbrange 275 290'
'set grads off'
'set clab forced'
'set parea 0.6 9.4 0.75 7.5'
'set gxout shaded'
'set csmooth on'
'set t 'i
'd temp'
'run cbarn.gs'
'set gxout contour'
'd temp'
'set gxout shaded'
drawSta1
'query dims'
str = sublinresult,5
date = subwrdstr,6
subtitl = 'Temperature K at 700mb - 'date''
'draw title 'subtitl
i = i + 1
'print'
'disable print'
endwhile
i = 1
while i = _nbFiles
'c'
if _hour = 0 | _hour = 4 & _type = 0
'enable print gh'i'_zoom2.meta'
else
'enable print gh'i'bis_zoom2.meta'
endif
'set rbrange 3050 3200'
'set grads off'
'set clab forced'
'set parea 0.6 9.4 0.75 7.5'
'set gxout shaded'
'set csmooth on'
'set t 'i
'd gh'
'run cbarn.gs'
'set gxout contour'
'd gh'
'set gxout shaded'
drawSta1
'query dims'
str = sublinresult,5
date = subwrdstr,6
subtitl = 'Geopotential height m at 700mb - 'date''
'draw title 'subtitl
i = i + 1
'print'
'disable print'
endwhile
i = 1
while i = _nbFiles
39
40
Meteorological Prediction Software: Maintenance Manual
'c'
if _hour = 0 | _hour = 4 & _type = 0
'enable print wind'i'_zoom2.meta'
else
'enable print wind'i'bis_zoom2.meta'
endif
'set rbrange 0 80'
'set lev 200'
'set grads off'
'set clab forced'
'set parea 0.6 9.4 0.75 7.5'
'set gxout stream'
'set csmooth on'
'set t 'i
'd uvmagu,v'
'run cbarn.gs'
drawSta1
'query dims'
str = sublinresult,5
date = subwrdstr,6
subtitl = 'Wind colored by mag. ms at 200mb - 'date''
'draw title 'subtitl
i = i + 1
'print'
'disable print'
endwhile
*** drawing of the curves time x mb
*** these functions use the bilinear interpolation
*** implemented in fortran dir $GADDIRudf$
'c'
'set t 1 '_nbFiles
'set lev 100 1000'
'set grads off'
'set lon -71'
'set lat -30'
'temp1 = temp'
'gh1 = gh'
'u1 = u'
'v1 = v'
'set lon -70.5'
'temp2 = temp'
'gh2 = gh'
'u2 = u'
'v2 = v'
'set lat -29.5'
'temp3 = temp'
'gh3 = gh'
'u3 = u'
'v3 = v'
'set lon -71'
'temp4 = temp'
'gh4 = gh'
'u4 = u'
'v4 = v'
'set yflip on'
'set zlog on'
if _hour = 0 | _hour = 4 & _type = 0
'enable print lasil_temp.meta'
else
'enable print lasil_temp2.meta'
endif
'd biliLaSitemp1,temp2,temp3,temp4'
'draw title Temperature K above La Silla 772 mB'
'run cbarn.gs'
'set gxout contour'
'd biliLaSitemp1,temp2,temp3,temp4'
'print'
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
'disable print'
'c'
'set t 1 '_nbFiles
'set lev 100 1000'
'set grads off'
'set yflip on'
'set zlog on'
if _hour = 0 | _hour = 4 & _type = 0
'enable print lasil_gh.meta'
else
'enable print lasil_gh2.meta'
endif
'set gxout shaded'
'd biliLaSigh1,gh2,gh3,gh4
'
'run cbarn.gs'
'draw title Geopotential Height m
above La Silla 772 mB
'
'set gxout contour'
'd biliLaSigh1,gh2,gh3,gh4
'
'print'
'disable print'
'c'
'set t 1 '_nbFiles
'set lev 100 1000'
'set grads off'
'set yflip on'
'set zlog on'
if _hour = 0 | _hour = 4 & _type = 0
'enable print lasil_modwin.meta'
else
'enable print lasil_modwin2.meta'
endif
'set gxout shaded'
'd magbiliLaSiu1,u2,u3,u4
,biliLaSiv1,v2,v3,v4
'
'run cbarn.gs'
'draw title Magnitude of wind ms
above La Silla 772 mB
'
'set gxout contour'
'd magbiliLaSiu1,u2,u3,u4
,biliLaSiv1,v2,v3,v4
'
'print'
'disable print'
'c'
'set t 1 '_nbFiles
'set lev 100 1000'
'set grads off'
'set yflip on'
'set zlog on'
if _hour = 0 | _hour = 4 & _type = 0
'enable print lasil_vecwin.meta'
else
'enable print lasil_vecwin2.meta'
endif
'set gxout vector'
'd biliLaSiu1,u2,u3,u4
biliLaSiv1,v2,v3,v4
'
'draw title Evolution of wind above La Silla 772 mB
'
'print'
'disable print'
'quit'
***
***
***
function drawStastate
'set line 1 1 1'
'set strsiz 0.1 0.1'
'set string 1 bl 3'
'q w2xy -70.'42*10060' -29.'16*10060
41
42
Meteorological Prediction Software: Maintenance Manual
x = subwrdresult,3
y = subwrdresult,6
'draw mark 3 'x' 'y' .1'
y = y + 0.1
x = x- 0.5
'draw string 'x' 'y' La Silla'
'q w2xy -70.'24*10060' -24.'37*10060
x = subwrdresult,3
y = subwrdresult,6
'draw mark 3 'x' 'y' .1'
y = y + 0.1
x = x - 0.5
'draw string 'x' 'y' Paranal'
'q w2xy -67.'45*10060' -23.'1*10060
x = subwrdresult,3
y = subwrdresult,6
'draw mark 3 'x' 'y' .1'
y = y + 0.1
x = x - 0.5
'draw string 'x' 'y' Chajnantor'
if state = 0
'query w2xy -71.5 -26.5'
xLo = subwrdresult,3
yLo = subwrdresult,6
'query w2xy -67 -22'
xHi = subwrdresult,3
yHi = subwrdresult,6
'draw rec 'xLo' 'yLo' 'xHi' 'yHi
'query w2xy -71.5 -31.5'
xLo = subwrdresult,3
yLo = subwrdresult,6
'query w2xy -67 -27'
xHi = subwrdresult,3
yHi = subwrdresult,6
'draw rec 'xLo' 'yLo' 'xHi' 'yHi
endif
8.3
The verication scripts verif0.gs, writetxt.gs
Two GrADS scripts are dedicated to the forecasts verication. The rst, verif0.gs, is called by the
UNIX-script step2 see section 7.1.3. It deals with the site of La Silla, as the local data for Paranal and
Chajnantor are not available in real time mode from Garching. It calls some user dened FORTRAN
functions biliLaSi and cuspLaSi , which will be described later in this maintenance manual.
The script is quite simple, it uses the four points nearest to the La Silla site to perform a bilinear
interpolation for the latitude and longitude, and it uses a cubic spline interpolation for the altitude of
the site.
When the local data for Paranal will be available, one just needs to update the script getlastmeteo in
the directory diskbasmprediVerifParanal and to add to the crontab le on olasg the same lines
than for La Silla.
The listing of verif0.gs follows :
verif0.gs
function mainargs
'reinit'
_day = subwrdargs,1
_hour = subwrdargs,2
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
43
_type = 0
file_ctl='ESD'_day''_hour'_3.ctl'
'open 'file_ctl
'set t 1 7'
'set lev 100 1000'
'set lon -70.5'
'set lat -29'
'temp1 = temp'
'gh1 = gh'
'u1 = u'
'v1 = v'
'set lat -29.5'
'temp2 = temp'
'gh2 = gh'
'u2 = u'
'v2 = v'
'set lon -70'
'temp3 = temp'
'gh3 = gh'
'u3 = u'
'v3 = v'
'set lat -29'
'temp4 = temp'
'gh4 = gh'
'u4 = u'
'v4 = v'
'd cuspLaSibiliLaSigh1,gh2,gh3,gh4,temp,biliLaSitemp1,temp2,temp3,temp4'
'd cuspLaSibiliLaSigh1,gh2,gh3,gh4,ws,
magbiliLaSiu1,u2,u3,u4,biliLaSiv1,v2,v3,v4'
'quit'
The second script, writetxt.gs, is dedicated to the java applet located in the verication pages on the
Web. It produces a data le in a format recognized by the java applet. For now, this applet only deals
with the La Silla site. It calls the writPred function, which is a FORTRAN user dened function for
GrADS.
In fact, as we will see later, it is a version slightly modied of the cubic spline interpolation program.
The GrADS script itself is quite simple and similar to the rst one, it only uses the four point nearing
La Silla and calls the writPred function.
The listing follows :
writetxt.gs
function mainargs
'reinit'
_day = subwrdargs,1
_hour = subwrdargs,2
_type = 4
file_ctl='ESD'_day''_hour'_3.ctl'
'open 'file_ctl
'set t 1 7'
'set lev 100 1000'
'set lon -70.5'
'set lat -29'
'temp1 = temp'
'gh1 = gh'
'u1 = u'
44
Meteorological Prediction Software: Maintenance Manual
'v1 = v'
'set lat -29.5'
'temp2 = temp'
'gh2 = gh'
'u2 = u'
'v2 = v'
'set lon -70'
'temp3 = temp'
'gh3 = gh'
'u3 = u'
'v3 = v'
'set lat -29'
'temp4 = temp'
'gh4 = gh'
'u4 = u'
'v4 = v'
'd writPredbiliLaSigh1,gh2,gh3,gh4,biliLaSitemp1,temp2,temp3,temp4,
magbiliLaSiu1,u2,u3,u4,biliLaSiv1,v2,v3,v4'
'close 1'
file_ctl='ESD'_day''_hour'_6.ctl'
'open 'file_ctl
'set t 1 17'
'set lev 100 1000'
'set lon -70.5'
'set lat -29'
'temp1 = temp'
'gh1 = gh'
'u1 = u'
'v1 = v'
'set lat -29.5'
'temp2 = temp'
'gh2 = gh'
'u2 = u'
'v2 = v'
'set lon -70'
'temp3 = temp'
'gh3 = gh'
'u3 = u'
'v3 = v'
'set lat -29'
'temp4 = temp'
'gh4 = gh'
'u4 = u'
'v4 = v'
'd writPredbiliLaSigh1,gh2,gh3,gh4,biliLaSitemp1,temp2,temp3,temp4,
magbiliLaSiu1,u2,u3,u4,biliLaSiv1,v2,v3,v4'
'quit'
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
45
9 Maintenance of the Fortran programs
9.1 The ECMextrDAT utility
This utility was coded in Fortran in order to decode the GRIB92 format used by ECMWF to disseminate its products ECMWF 97. The source code is located on olasg, in the home2 asmpredi ECMWF software
directory. In this directory must be located the two librairies pbiolib.a and gridlib.a needed by
ECMextrDAT for the compilation, the makele and the source le ECMextrDAT.f. The source le
being quite long, it will not be printed in the manual. Here is the makele for this utility :
#
#
configuration file for hp
#
CC
= cc
CFLAGS
= -g -Aa -D_INCLUDE_HPUX_SOURCE -D_INCLUDE_XOPEN_SOURCE -D_INCLUDE_AES_SOURCE -D_INCLUDE_POSIX_SOURCE -D_
XPG2 -Dhp
FC
= f77
FFLAGS
= -g +U77 -v -WF,"-P" +z -Dhp
RANLIB
=
CT = bintrue
RANLIB = bintrue
.SUFFIXES:
.f
.f.o:
$FC $FFLAGS -c $
GRIBLIB = griblib.a
PBIOLIB = pbiolib.a
PROGS = ECMextrDAT
PROGS_OBJ = ECMextrDAT.o
all
: $PROGS
$PROGS : $PROGS_OBJ
$FC $FFLAGS [email protected] $GRIBLIB $PBIOLIB -o [email protected]
clean :
rm -f $SOURCES
The aim of this utility is to produce some les understandable by the GrADS utility. Then it must
produce binary les without headers for the data, and text les the ctl les to inform GrADS about
the content of those binary les. To produce the binary les, ECMextrDAT must read all the les
from ECMWF composing a set of prediction les, and sort the information before recording it in new
binary les. The challenge is that the les from ECMWF are sorted by time, and not in geographical
region as they should be. Another problem comes from the fact that the products are not ordered
in a logical manner. Sometimes the le uses the sequence 'general-paranal-la silla' and on another
occasion is 'general-la silla-paranal'.
So the ECMextrDAT utility looks for patterns in fact, those patterns are the coordinates of the origin
point of the grid, the type of data and the level of forecast in the headers to identify the geographical
regions, the time and the level. It lls an array with the index of the records. It calls then a procedure
to sort out this array, to order the records. It then reloads all the records and produces the three les
needed by GrADS. The sensitive point is the pattern that the program is looking for in the ECMWF
headers. If the patterns change ECMWF could decide to use another coding, one must re-enter the
46
Meteorological Prediction Software: Maintenance Manual
new patterns in the program for it to work properly. If it is not the case the program will produce
corrupted les for GrADS.
9.2 The user-dened functions for GrADS
9.2.1 The bilinear interpolation biliLaSi, biliPara, biliChaj
The rst user-dened function UDF for the GrADS utility is a small UDF which performs bilinear
interpolation in latitude and longitude for a given point. The present version of GrADS does not allow
to pass to an UDF the number of arguments we want. For instance we could not pass the the coordinates of the four points of the grid nearing the considered point. For this reason, we had to develop three
versions of this utility, one for each point of interest. In the directory diskbasmprediGrADSudf,
one could nd three utility for the interpolation, biliPara, biliLaSi and biliChaj.
Those UDF are quite simple to understand. They take from a temporary le written by GrADS
the data concerning the four points, and perform the bilinear interpolation and write the result in a
temporary le read by GrADS. The listing of the biliLaSi utility is given here :
c
c
c
PROGRAM BiliLaSi
integer nbTime
real vals20,ovals20
real dummy9
real x1153,x2153,x3153,x4153
real result153
c
c
c
open 8,file = 'diskbasmprediGrADSudfbiliLaSi.in',
form = 'unformatted'
x
c
c
read 8
read 8 vals
nbTime = vals5
c
c
read 8 x1i,i=1,9*nbTime
read8 dummy
read 8
read 8 vals
read 8 x2i,i=1,9*nbTime
read8 dummy
c
read 8
read 8 vals
read 8 x3i,i=1,9*nbTime
read8 dummy
c
read 8
read 8 vals
read 8 x4i,i=1,9*nbTime
read8 dummy
c
5
c
do 5 i = 1,9*nbTime
resulti = interLx1i,x2i,x3i,x4i
continue
ovals1 = 0.0
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
47
open 10,file = 'diskbasmprediGrADSudfbiliLaSi.out',
form = 'unformatted'
write10
ovals
write10
vals
write10
resulti
,i=1,9*nbTime
write10
dummy
close 8
close 10
end
x
c
c
c
subroutine interLval1,val2,val3,val4
c
real val1,val2,val3,val4
c
c
c
real t,u,result
t = -70.7 + 71
-70.5 + 71
u = -29.7 + 30
-29.5 + 30
result
x
+
x
+
x
+
c
= 1 - t
* 1 - u
* val1
t * 1 - u
* val2
t * u * val3
1 - t
* u * val4
return result
end
To modify this script for a fourth site, one must notice that the only things to change are the coordinates of the nearest points in the grid, and the coordinates of the site.
9.2.2 The cubic spline interpolation cuspLaSi, cuspPara and cuspChaj
Those UDF which are again three, one for each site, perform a double cublic spline interpolation on
the altitude in order to extract the best approximation of the considered parameter at the site ground
level. The problem is that we do not have the parameter as a function of the altitude but standard
level. The solution was then to call a rst time the cubic spline procedure on the geopotential height
parameter gives a correlation between the altitude and the pressure at the altitude of the site, which
delivers the pressure value, and then to call again the cubic spline interpolation at this pressure value
on the requested parameter to extract the best approximation of this parameter.
9.2.3 The writPred and writPre2 functions
This user dened function, located on olasg, is used to produce one of the data le for the java applet
in the verication pages. It is a GrADS user function because it needs to have access to the ECMWF
predictions. In fact it opens the predictions given by GrADS, it processes them through a Kalman
lter and then writes a report le lasi.txt or Para.txt in the directories diskbasmprediVerifLaSilla
or diskbasmprediVerifParanal.
The second function, writPre2, is the same as the rst but deals with the Paranal site.
The sources are in the directory diskbasmprediGrADSudf.
48
Meteorological Prediction Software: Maintenance Manual
10 Maintenance of other programs
10.1 The IDL programs
The main IDL program being used is the one which deals with the satellite images,TryRead.pro. It
is located on the olasg machine, in the directory home2asmprediCIRAutils. It is launched by
the script satellite which calls the IDL initialization le, satellite.pro, and which launches the main
program with the correct lename.
The crontab line for this script follows:
00 * * * * bintcsh home2asmprediCIRAutilssatellite devnull 2&1
Basically, it is based on the function Gview.pro developped by CIRA. We have taken the core program
of the function the reading and the decoding of a satellite image le, and added some functionalities,
for example writing the name of the three sites or the date and the hour at which the image was taken.
This application was developped using the version 5.1 of IDL, installed on olasg. This installation
hass a single user license.
10.2 The Java applets
Two applet have been developped for the Web astro-climatology site. The rst one deals with the
satellite images, and allows the user to browse through the last 8 images. The java denition le
is named Satellite.java, the binary version being Satellite.class. This applet needs 16 image les to
function properly 8 les clouds x.gif and 8 les zoom x.gif . In order to produce the animation loop
while being able to interact with the user interface, we needed to use a special thread for the animation,
and that is the only diculty of the applet.
The second applet is behind the link verication pages, and allows the user to extract the corrected
forecasts for the next 8 days. It reads on the server web1 and web2 in this case three URL's
corresponding to the three data les for the applet lasi.txt, para.txt and chaj.txt . If the les are not
present, the applet will throw an exception but will continue to execute properly it is due to the fact
that the ground data for Paranal and Chajnantor are not available. The source le for this applet is
Forecast.java, the binary le being Forecast.class.
Those two applets were developed with the JDK 1.0, to ensure their compatibility with most of the
recent browsers i.e. Netscape 3.0 or later and Internet Explorer 3.0 or later. One could port them
to JDK 1.1 in the future, but as they are only basical applets there is no obligation to do so.
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
49
References
CRS4 96 R. Benzi,R. Deidda, M. Marrocu, A. Speranza, Feasibility study of a meteorological
prediction model for ESO observatories in Chile Phases A1,A2, 1996 ESO contract number
45060VLT956952GWI
CRS4 97 R. Benzi,R. Deidda, M. Marrocu, A. Speranza, Feasibility study of a meteorological
prediction model for ESO observatories in Chile Phases A3,A4, 1997 ESO contract number
45060VLT956952GWI
ECMWF 95 ECMWF, User Guide to ECMWF Products 2.1, 1995
ECMWF 97 D. Jokic, The dissemination of ECMWF products to Member States, 1997
Press et al. 94 W. Press, S. Teulkosky, W. Vetterling, B. Flannery, Numerical Recipes in FORTRAN
second edition, 1994
50
Meteorological Prediction Software: Maintenance Manual
A Appendix
A.1 Location of the Unix-scripts and their related les
This table summarizes the location of the UNIX-scripts and their related les, and shows the use of
each of the le. One should notice that each script le uses some temporary les tempx which are
not mentionned here because of their temporary existence.
File
ECMtoOlasg
Machine
Location
opus3 account homeasmecmwf
asmpredi
ECMtoOlasg2
opus3 account homeasmecmwf
asmpredi
scrWeb0
olasg
home2asmprediECMWFdatatmp
scrWeb4
olasg
home2asmprediECMWFdatatmp
day0.txt
olasg
home2asmprediECMWFdatatmp
day4.txt
olasg
home2asmprediECMWFdatatmp
ECMWF.result olasg
diskbasmprediVerif
satellite
olasg
home2asmprediCIRAutils
satellite.report
last
olasg
olasg
diskbasmprediVerif
home2asmprediCIRAutils
current
olasg
home2asmprediCIRAutils
getlastmeteo
olasg
diskbasmprediVerifLaSilla
step1
olasg
diskbasmprediVerif
step2
olasg
diskbasmprediVerif
step3
olasg
diskbasmprediVerif
Use
transfers complete set of
ECMWF les product of
12UT
transfers complete set of
ECMWF les product of
00UT
process set of 00UT les from
ECMWF to produce images
for the Web pages
process set of 12UT les from
ECMWF to produce images
for the Web pages
contains the last day for
which the 00UT product has
been processed
contains the last day for
which the 12UT product has
been processed
'log' le for the treatment of
the ECMWF les
upload the most recent satellite image from CIRA and
process it through an IDL
script
'log' le for the satellite script
text le containing the name
of the last uploaded image
text le containing the name
of the current transfer, if
there is any
script to extract some data
from the base archeso
rst part in the verication
process measures
second part in the verication
process forecasts
third part in the verication process kalman corrected forecasts
51
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
File
te.log and te.log.bak
Machine
olasg
Location
diskbasmprediVerif
input.te
olasg
diskbasmprediVerif
pr.log and pr.log.bak
olasg
diskbasmprediVerif
input.pr
olasg
diskbasmprediVerif
ws.log and ws.log.bak olasg
diskbasmprediVerif
input.ws
olasg
diskbasmprediVerif
publish0
web4
disksdiskwawebdocsesonewgen-fac
pubsastclimforecastmeteoscripts
publish4
web4
disksdiskwawebdocsesonewgen-fac
pubsastclimforecastmeteoscripts
verif
web4
data.txt
olasg
disksdiskwawebdocsesonewgen-fac
pubsastclimforecastmeteoscripts
diskbasmprediVerif
goesGetSat
web4
cloudsx.gif
olasg
disksdiskwawebdocsesonewgen-fac
pubsastclimforecastmeteoscripts
home2asmprediCIRAutils
zoomx.gif
olasg
home2asmprediCIRAutils
Use
log le and his backup for
the temperature
input le temperature for
gnuplot
log le and his backup for
the pressure
input le pressure for
gnuplot
log le and his backup for
the wind speed
input le wind speed for
gnuplot
uploads the images from
olasg to the Web pages
product of 12UT
uploads the images from
olasg to the Web pages
product of 00UT
uploads the verication images to the Web pages
text le containing some information for the java applet
on the verication pages
uploads the satellite images
and publish them to the web
post-processed
satellite
images
corresponding
zoom
of the post-processed satellite
images
52
Meteorological Prediction Software: Maintenance Manual
A.2 Location of the GrADS applications and their related les
In this table are presented the name of the GrADS scripts and utilities, and their dependent les.
File
mainMeteo.gs
Machine
olasg
Location
home2asmprediECMWFdatatmp
cbarn.gs
olasg
home2asmprediECMWFdatatmp
usr colors.gs
usr sta.gs
and olasg
home2asmprediECMWFdatatmp
par draw.5.gs
par bar.5.gs
and olasg
home2asmprediECMWFdatatmp
script1.gs
olasg
home2asmprediECMWFdatatmp
script2.gs
olasg
home2asmprediECMWFdatatmp
script3.gs
olasg
home2asmprediECMWFdatatmp
and olasg
home2asmprediECMWFdatatmp
tempx.gif
tempx bis.gif
ghx.gif and ghx bis.gif
olasg
home2asmprediECMWFdatatmp
and olasg
home2asmprediECMWFdatatmp
tempx zoom1.gif and olasg
tempx bis zoom1.gif
home2asmprediECMWFdatatmp
ghx zoom1.gif
and olasg
ghx bis zoom1.gif
home2asmprediECMWFdatatmp
windx zoom1.gif and olasg
windx bis zoom1.gif
home2asmprediECMWFdatatmp
tempx zoom2.gif and olasg
tempx bis zoom2.gif
home2asmprediECMWFdatatmp
ghx zoom2.gif
and olasg
ghx bis zoom2.gif
home2asmprediECMWFdatatmp
windx.gif
windx bis.gif
Use
Front-end utility for manipulating the data les from
ECMWF
Utility to draw the colour
scale next to the GrADS
graphics
GrADS data les. The rst
contains the user dened colors, the second the coordinates of the stations to draw
on the maps
GrADS data les. They contain the parameters and the
buttons to draw in the frontend utility
called by scrWebx. Produces
the images for the generic
region
called by scrWebx. Produces
the images for the Paranal
region
called by scrWebx. Produces
the images for the La Silla
region
gif of the temperature for the
generic region - short term
and long term
gif of the geopotential height
for the generic region - short
term and long term
gif of the wind streamlines
for the generic region - short
term and long term
gif of the temperature for the
Paranal region - short term
and long term
gif of the geopotential height
for the Paranal region - short
term and long term
gif of the wind streamlines
for the Paranal region - short
term and long term
gif of the temperature for the
La Silla region - short term
and long term
gif of the geopotential height
for the La Silla region - short
term and long term
53
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
File
Machine
windx zoom2.gif and olasg
windx bis zoom2.gif
Location
home2asmprediECMWFdatatmp
xxxxx
xxxxx
temp.gif
temp2.gif
and olasg
home2asmprediECMWFdatatmp
xxxxx
xxxxx
gh.gif
gh2.gif
and olasg
home2asmprediECMWFdatatmp
xxxxx
xxxxx
vecwin.gif and olasg
vecwin2.gif
home2asmprediECMWFdatatmp
xxxxx
xxxxx
modwin.gif and olasg
modwin2.gif
home2asmprediECMWFdatatmp
verif0.gs
olasg
diskbasmprediVerif
writetxt.gs
olasg
diskbasmprediVerif
Use
gif of the wind streamlines
for the La Silla region - short
term and long term
gif of the reconstructed radio
prole of temperature for site
xxxxx chajn, lasil or paranshort term and long term
gif of the reconstructed radio
prole of geopotential height
for site xxxxx chajn, lasil or
paran- short term and long
term
gif of the reconstructed radio
prole of the directions of the
wind vectors for site xxxxx
chajn, lasil or paran- short
term and long term
gif of the reconstructed radio
prole of wind speed for site
xxxxx chajn, lasil or paranshort term and long term
updates the three log les for
the verication. Called by
the 'step2' script
writes the text le for the
java applet in the verication
pages
54
Meteorological Prediction Software: Maintenance Manual
A.3 Location of the Fortran programs
In this table are presented the fortran programs used by the utilities.
File
ECMextrDAT
Machine
olasg
Location
home2asmprediECMWFsoftware
ECMextrDAT.f
olasg
home2asmprediECMWFsoftware
griblib.a and pbiolib.a
olasg
home2asmprediECMWFsoftware
biliLaSi biliLaSi.f
olasg
diskbasmprediGrADSudf
biliPara biliPara.f
olasg
diskbasmprediGrADSudf
biliChaj biliChaj.f
olasg
diskbasmprediGrADSudf
cuspLaSi cuspLaSi.f
olasg
diskbasmprediGrADSudf
cuspPara cuspPara.f
olasg
diskbasmprediGrADSudf
cuspChaj cuspChaj.f
olasg
diskbasmprediGrADSudf
writPred writPred.f
olasg
diskbasmprediGrADSudf
writPre2 writPre2.f
olasg
diskbasmprediGrADSudf
Use
utility to decode the les from
ECMWF
source le of the ECMextrDAT utility
librairies needed to compile
the ECMextrDAT utility
bilinear interpolation for the
La Silla site
bilinear interpolation for the
Paranal site
bilinear interpolation for the
Chajnantor site
cubic spline interpolation for
the La Silla site
cubic spline interpolation for
the Paranalsite
cubic spline interpolation for
the Chajnantor site
kalman ltering of the predictions for the java verication
applet for the La Silla site
kalman ltering of the predictions for the java verication
applet for the Paranal site
55
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
A.4 Location of the IDL applications and their related les
In this table are presented the IDL applications and their related les.
File
satellite.pro
Machine
olasg
Location
home2asmprediCIRAutils
TryRead.pro
olasg
home2asmprediCIRAutils
namele
olasg
home2asmprediCIRAutils
current
olasg
home2asmprediCIRAutils
last
olasg
home2asmprediCIRAutils
cloudsx.gif
olasg
home2asmprediCIRAutils
zoomx.gif
olasg
home2asmprediCIRAutils
Use
header le used to execute
the main IDL applicationand
to set some global variables
main IDL application to read
and produce the gif les of
the satellite images
text le containing the name
of the satellite image being
processed by IDL
text le containing the name
of
the
satellite
image le being downloaded
fron CIRA present only during a download
text le containing the name
of the last succesfully transferred satellite image le
processed image le of the
general region, 2 1 8
processed image le of the
zoomed region, 2 1 8
x
::
x
::
56
Meteorological Prediction Software: Maintenance Manual
A.5 Location of the java applets and their related les
In this table are presented the java applets and their related les.
File
Machine
Sat anim.class
web4
Forecasts.java
web4
Forecasts.class
web4
cloudsx.gif
web4
zoomx.gif
web4
lasi.txt
web4
para.txt
web4
chaj.txt
web4
Sat anim.java
web4
Location
Use
disksdiskwawebdocsesonewgen-fac source le of the satellite
pubsastclimforecastmeteoCIRA
applet
disksdiskwawebdocsesonewgen-fac binary le of the satellite
pubsastclimforecastmeteoCIRA
applet
disksdiskwawebdocsesonewgen-fac source le of the verication
pubsastclimforecastmeteoverication applet
disksdiskwawebdocsesonewgen-fac binary le of the verication
pubsastclimforecastmeteoverication applet
disksdiskwawebdocsesonewgen-fac processed image le of the
pubsastclimforecastmeteoCIRAimagesgeneral region, 2 1 8
disksdiskwawebdocsesonewgen-fac processed image le of the
pubsastclimforecastmeteoCIRAimageszoomed region, 2 1 8
disksdiskwawebdocsesonewgen-fac text le for the verication
pubsastclimforecastmeteoverication applet, La Silla site
disksdiskwawebdocsesonewgen-fac text le for the verication
pubsastclimforecastmeteoverication applet, Paranal site
disksdiskwawebdocsesonewgen-fac text le for the verication
pubsastclimforecastmeteoverication applet, Chajnantor site
x
::
x
::
VLT-TRE-ESO-17443-1678, P. Labit, Oct. 19, 1998
B List of crontab les
In this appendix are presented the crontab les present on the machines.
Machine opus3, account asmpredi
00,20,40 * * * * homeasmecmwfECMtoOlasg devnull 2&1
10,30,50 * * * * homeasmecmwfECMtoOlasg2 devnull 2&1
Machine olasg, account asmpredi
00,30 * * * * home2asmprediECMWFdatatmpscrWeb0 devnull 2&1
15,45 * * * * home2asmprediECMWFdatatmpscrWeb4 devnull 2&1
00 * * * * bintcsh home2asmprediCIRAutilssatellite devnull 2&1
00 21 * * * diskbasmprediVerifLaSillastep1 devnull 2&1
00 04,09 * * * diskbasmprediVerifLaSillastep2 devnull 2&1
Machine web4, account msarazin
00 * * * * disksdiskwawebdocsesonewgen-facpubsastclimforecast
meteoscriptspublish4 devnull 2&1
00 * * * * disksdiskwawebdocsesonewgen-facpubsastclimforecast
meteoscriptspublish4 devnull 2&1
00 * * * * disksdiskwawebdocsesonewgen-facpubsastclimforecast
meteoscriptspublish0 devnull 2&1
30 01,04,07,10,13,16,19,22 * * * disksdiskwawebdocsesonewgen-fac
pubsastclimforecastmeteoscriptsgoesGetSat devnull 2&1
00 05 * * * disksdiskwawebdocsesonewgen-facpubsastclimforecast
meteoscriptsverif devnull 2&1
57